The SFM places each transaction it receives into one of its log files. For the transaction to be accepted and written into the transaction log file, a transaction must pass certain tests. If it does not pass these tests, the SFM writes the transaction to its unrouteable log file.
The user who has access to the unrouteable log file
through TRAN-IDE can correct these transactions and re-submit them
to the SFM for processing.
The SFM saves transactions that it cannot post to destinations in the transaction log file. If one or more production objects process a transaction before it goes to its destinations, the SFM saves the original transaction until the destination is available, then processes the transaction through its production objects and sends it on.
When the SFM receives notification that a transaction in the transaction log file has been received and processed by its destination delivery AIM, the SFM marks the transaction as processed. If the incoming transaction activity is high enough so that all of the file’s space is used up, the SFM wraps to the beginning of the file and reuses the space used by processed transactions.
The SFM uses a simple transaction ID or production ID value to route a transaction through the SFM distributed computing environment. You must assign an ID value to each unique transaction record structure.
A production ID value is the name of a production object generated with TRAN-IDE. You may use one of the routing functions described in this chapter to send a transaction directly for processing by a single production object. In this case, SFM does not check the transaction to see if it qualifies for any other production objects.
| Copyright © 2005. Sybase Inc. All rights reserved. |
|
|
